Related Experiment Video
Updated: Jun 19, 2026
11:54
Microfluidic Platform with Multiplexed Electronic Detection for Spatial Tracking of Particles
Published on: March 13, 2017
Scalable multiplexed machine learning gas sensor chips for food classification
Carla Bassil1,2,3, Kichul Lee1,4, Xun Liao1,2
1Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, CA, USA.
Science Advances
|June 17, 2026
Summary
A novel 16-element gas sensor chip with distinct layers enables scent-based object identification. This heterogeneous array, using carbon nanotube transistors and machine learning, achieved 92.6% accuracy in classifying diverse items like allergens.
Area of Science:
- Materials Science
- Sensor Technology
- Machine Learning
Background:
- Multiplexed gas sensor arrays and machine learning advance scent-based sensing.
- Existing platforms face limitations due to material overlap and complex fabrication.
Purpose of the Study:
- To develop a novel 16-element monolithic gas sensor chip with distinct sensing layers.
- To enable scalable, heterogeneous sensor arrays for scent identification.
Main Methods:
- Fabrication of a monolithic chip with 16 distinct sensing layers using carbon nanotube field-effect transistors.
- A single-step microdispensing method compatible with automated pipetting for functionalization.
- Application of machine learning algorithms for automated object identification based on scent profiles.
Main Results:
- The developed chip demonstrated characteristic signal patterns for specific scent profiles.
- Automated classification of 16 different objects, including food spoilage and nut allergens.
- Achieved an overall prediction accuracy of 92.6% for object identification.
Conclusions:
- The heterogeneous sensor array offers a scalable solution for advanced scent-based sensing.
- The system effectively identifies objects through scent analysis using machine learning.
- This technology has potential applications in areas like food safety and allergen detection.
